CHAPTER 3: THE DISCRETE –TIME FOURIER ANALYSIS

CHAPTER OBJECTIVES
Upon a successful completion of this chapter, the student will be able to:

Understand the properties of the Discrete Time Fourier Transform (DTFT).

Represent the LTI System in the Frequency Domain.
Understand the effect in the frequency domain of sampling a given signal
in the time domain.
Reconstruct the continuous time signal from its samples using ideal lowpass
filter.
LECTURE # 4
CHAPTER OUTLINES
3.1 DISCRETE –TIME FOURIER TRANSFORM (DTFT)
3.1.1 PROPERTIES OF DTFT
3.1.1.1 PERIODICITY
3.1.1.2 LINEARITY
3.1.1.3 TIME SHIFTING
3.1.1.4 FREQUENCY SHIFTING
3.1.1.5 SYMMETRY
3.1.1.6 FOLDING
3.1.1.7 CONVOLUTION
3.1.1.8 MULTIPLICATION
3.1.1.9 ENERGY Lecture # 4 1
 EXAMPLE 3.1
3.2 Frequency Domain Representation of LTI System
EXAMPLE 3.2

LECTURE # 5
3.3 RELATIONSHIP BETWEEN FREQUENCY RESPONSE AND DIFFERENCE
EQUATION FOR LTIS
EXAMPLE 3.3
EXAMPLE 3.4

3.4 SAMPLING OF ANALOG SIGNALS

3.4 SAMPLING THEOREM

Lecture # 4 2
3.1 DISCRETE –TIME FOURIER TRANSFORM (DTFT)

The discrete-time Fourier Transform

 xn    


xe   xn   xne


jw  jwn

n   n  

The Inverse discrete-time Fourier Transform (IDTFT)

xn     X e  
1 
 X e .e dw
1 jw jw jwn

2 

Notes:
• X(ejw) is a complex valued continuous function

• w = 2π f [rad/sec]

• f is the digital frequency measured in [ C/S]

Lecture # 4 3
 3.1.1 PROPERTIES OF DTFT

3.1.1.1 PERIODICITY
X e jw
 Is periodic with period 2π

         
X e jw  X e j w2   X e jwe j 2  X e jw .1  X e jw

Therefore, we need for analysis only period of X(ejw) for w  0,2 

w    ,  

Not for the whole domain   w 

Lecture # 4 4
 3.1.1 PROPERTIES OF DTFT (Contd.)
3.1.1.2 LINEARITY
 
 x1 n  x2 n  x1 n  x2 n

3.1.1.3 TIME SHIFTING

A shift in the time domain corresponds to the phase shifting

xn  k   X e  e
jw  jwk

3.1.1.4 FREQUENCY SHIFTING

Multiplication by a complex exponential corresponds to a shift in the frequency domain

  
 xne jwo n  X e j wwo  
Lecture # 4 5
 3.1.1 PROPERTIES OF DTFT (Contd.)
3.1.1.5 SYMMETRY
For real-valued   is conjugate symmetric
xn, X e jw
   
X e jw  X  e jw
ReX e   ReX e 
jw  jw Even Symmetry

ImX e    ImX e 
jw  jw Odd Symmetry

   
X e jw  X e  jw Even Symmetry

X e   X e 
jw  jw
Odd Symmetry

Therefore, to plot X(ejw) we need to consider a half period of X(ejw)

Generally, in practice this period is chosen to be w  0,   

Lecture # 4 6
 3.1.1 PROPERTIES OF DTFT (Contd.)
3.1.16 FOLDING
Folding in the time domain corresponds to the folding in the frequency domain

 
x n  X e jw

3.1.1.7 CONVOLUTION
This is one of the most useful properties that make the system convenient for analysis

x1 n x2 n  x1 n.x2 n  X1 e  X e 

jw
2
jw

x(n) h(n) yn  xn hn

Time Domain X e  jw
H(ejw)
     
Y e jw  X e jw H e jw

Frequency Domain
Lecture # 4 7
 3.1.1 PROPERTIES OF DTFT (Contd.)
3.1.1.8 MULTIPLICATION
This is dual to the convolution property
x1 n.x2 n  x1 n x2 n

3.1.1.9 ENERGY
The energy of the sequence x(n) can be written as :


 X e 

E   xn 
1 jw 2

2
dw
 2 
This is known as the Parseval’s Theorem

Lecture # 4 8
 EXAMPLE 3.1
Consider the signal xn   0.5 U n
n

a- Determine its DTFT.

b- Evaluate X(ejw) at w = 0, 0.25π, 0.5π , 0.75 π and π
c- Plot its magnitude , angle, real part and imaginary part

Solution:
a- Since the signal x(n) is absolutely summable, therefore, DTFT exists:

    xne
 
 jwn
  0.5 e  jwn
jw n
X e
n   0

 0.5  e 

 jw n 1

0 
1  0.5  e  jw 
cosw  j sin w cosw  j sin w
X e   jw
e jw
jw
 
e  0.5 cosw  j sin w  0.5 cosw  0.5  j sin w
Lecture # 4 9
 
X e jw
 cosw  j sin w cosw  0.5  j sin w
cosw  0.52  sinw2

X e  
jw cosw  0.5cosw  sinw  jsinwcosw  sinwcosw  0.5 sinw
2 2

cosw  sinw  0.25  cosw

2 2

X e  
jw 1  0.5 cosw  j 0.5 sin w
1.25  cosw
1  0.5 cosw
 
X e jw 
1  0.5 cosw2  0.5 sinw2   
Re X e jw 
1.25  cosw
1.25  cosw2
 0.5 sin w
 
X e jw  tan 1
 0.5 sin w   
Im X e jw

1.25  cosw
1  0.5 cosw

Lecture # 4 10
 Magnitude Response
Magnetude Part Real PartPart
Magnetude
2 2

1.8 1.8

1.6 1.6

Magnetude
1.4
Magnetude

1.4

1.2
1.2

1
1

0.8
0.8

0.6
0.6 0 0.5 1 1.5 2 2.5 3 3.5
0 0.5 1 1.5
Frequency 2
[rad\sec] 2.5 3 3.5 Frequency
Frequency in Pi units
[rad/sec]
frequency in pi unitsude Imaginary
Phase Response
Angle Part ImaginaryPart
Part
0 0

-0.1 -0.1

-0.2 -0.2
Magnetude

Magnetude
-0.3 -0.3

-0.4 -0.4

-0.5 -0.5

-0.6 -0.6

-0.7 Lecture # -0.7

4 11
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
frequency in pi unitsude
Frequency [rad/sec] frequency in[rad/sec]
Frequency pi unitsude
3.2 Frequency Domain Representation of LTI System
The discrete-time Fourier transform of an impulse response is called
the frequency response or Transfer function of a LTI system

 n  hn 
H e 
LTI System
jw
1

   hne

 jwn
H e jw 
n  
For an arbitrary summable sequence x(n) as an input with impulse response h(n)

xn  hn 
yn  xn hn
 
X e jw
He jw      
Y e jw  X e jw  H e jw

Hint: yn   Y e 
1 jw
Lecture # 4 12
3.2 Frequency Domain Representation of LTI System (Contd.)
Special Cases :
Case 1:
If xn  Ae jwo n complex exponential


yn  A H e e  jwo j wo n  H ( e jwo ) 
Case 2:

If xn  A coswo n    Sinusoidal input


yn  A H e   cosw n    H e 
jwo
0
jwo

This is called the steady state response of the system yss n 

Lecture # 4 13
EXAMPLE 3.2

hn  0.9 U n

n
For the system described by its impulse response
Do the following
a- Determine the frequency response
b- Plot the magnitude and phase responses
c- Calculate the steady state response for the input xn  0.1U n
Solution

    hne
a- Using DTFT we find :  
  0.9 e  jwn
 jwn
jw n
He
1 n n 0

 0.9e 

 jw n 1
  
n   1  0.9e  jw
1  0.9 cosw  j 0.9sin w

He 
jw 1
1  0.9 cosw  0.9 sinw
Hence 2 2

Lecture # 4 14
 H e jw    1

1
1  0.9 cosw2  0.9 sinw2 
1  2  0.9 cosw  0.9  cosw  sin w
2 2 2

 
H e jw 
1
1.81  1.8  cosw


 0.9 sin w 

and
 
H e jw   arctan  
1  0.9 cosw
b- The magnitude and phase plots
Magnitude Response
Phase Response 10
0
9
-0.05
8

-0.1
7

6
Phase in Radians

-0.15

|H|
5
-0.2
4
-0.25
3

-0.3
2

-0.35 1

0
-0.4 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Frequency in pi units
Frequency in pi units

Lecture # 4 15
The steady state response [ at A = 0.1, θ = 0, wo = 0] is given by :

  
yss n  A H e jw cos wo n    H e jwo  

  
yss n  0.1 H e cos 0  n  0  H e
0
 
jw0

 
yss n  0.1 H e0  0.110  1

Lecture # 4 16
 3.3 RELATIONSHIP BETWEEN FREQUENCY RESPONSE
AND DIFFERENCE EQUATION FOR LTIS
When a LTIS is represented by the Difference Equation (DE) :
M N
yn    bi xn  i    ak yn  k 
i 0 k 1
Take the DTFT for both sides :

    b X e  e  e
M N
Ye jw
i
jw  jwi
  ak Y e jw  jwk

i 0 k 1
 
   
N M
Y e jw 1   ak e  jwk   X e jw  bi  e  jwi
 k 1  M
i 0

   a e  jwi

 
jw i
Ye
He  jw
 i 0

X e jw   N
1   bk e  jwk
k 1
Lecture # 5 17
 EXAMPLE 3.3
A LTI system is specified by the difference equation as:

yn  0.8yn 1  xn

a- Determine |H(ejw)|
b- Calculate the steady state response yss (n) that corresponds to the input x(n)

xn  cos0.05nU n

Solution :

a- Take the DTFT for both sides or apply the above equation:

   
Y e jw  0.8Y e jw e jw  X e jw  
H e  
jw Y e jw
 
1
X e  1  0.8e  jw
jw

Lecture # 5 18
b- When xn  cos0.05n
wo  0.05 and   0

  
yn  A H e jw cos wo n    H e jwo  
at wo = 0.05 π

H e  
1  j 0.5377
jw
 4.0928e
1  0.08e  j 0.05
Therefore,
yn  4.0928 cos0.05n  0.5377
 4.0928 cos0.05 n  3.42
This means that at the output, the sinusoidal is scaled by 4.0928 and

shifted by 3.42 samples

Lecture # 5 19
 EXAMPLE 3.4
A third order low pass filter is described by the difference equation:
yn  0.081xn  0.0543xn 1  0.0543xn  2  0.0181xn  3
 1.76 yn 1 1.1829 yn  2  0.27814 yn  3
a- Determine H(ejw)
b- Plot the magnitude and phase response of the filter.
c- Calculate the steady state response yss(n) that corresponds to the
input x(n)= 2 cos (0.5π n- π/4)
Solution Apply the above procedures as in Example 3.3 .
Magnitude Response Phase Response
1.4 4

1.2 3

2
1

Phase in Radians
1
0.8
|H|

0
0.6
-1

0.4
-2

0.2 -3

0 -4
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Lecture # 5 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 20
0.9 1
Frequency in pi units Frequency in pi units
 3.4 SAMPLING OF ANALOG SIGNALS
• The sequence x(nT) is obtained from a continuous time signal x(t) by sampling
• This is done multiplying a periodic impulse train s(t) [sampling function] by x(t)
• The period Ts is called the sampling period and Fs =1/Ts is the sampling frequency

ws  2Fs
xs t   st .xt 
xt  
st     t  nT  s
n   Sampling Function
st  t
st 
Ts
xt  xs t 
xs nTs  t

0 Ts 2Ts 3Ts 4Ts 5Ts …. nTs Lecture # 5 21

 Therefore, the sampled signal is given by :

xs t    xnT  t  nT 
n  

From the multiplicative property, X s w is the convolution of

X  jw and   jw

Therefore,

X s  jw    X  j S  jw   d
1
2 
It is known that
2 
S  jw     w  kw  s
Ts k  

Since, the convolution with an impulse simply shifts a signal


 X s  jw    X  jw  kw 
1
X  jw  w  wo   X  j w  wo  s
Ts k  
Lecture # 5 22
 
X s  jw    X  jw  kw 
1
s
Ts k  
That is X s  jw  Is a periodic function of w consisting of :
1
superposition of shifted replicas of X s  jw  scaled by
Ts
ws  2wM  ws  wM  wM ws  2wM  ws  wM  wM
X  jw X  jw
1 1

.
w
.0
.
w
.
wM
.0
.
w M
M M

S  jw  S  jw 
2 2

.
 2w
. Ts
.0 . . . . . . . . . . . .w
Ts
 4ws  3ws  2ws  ws 0 ws 2ws 3ws 4ws
s ws
w
s
2w s w
X s  jw  1
X s  jw 
1 Ts .
Ts
.  .2w . . .w . w. .0 w. . w. . . 2.w . . .4.w . . . . . w. ...0 w... . . . . . .
s s M
M s s w s M
M
w
ws  wM
w
Lecture # 5 ws  wMs 23
ws  wM ws  wM
 If ws  2wM The signal can be recovered exactly from xs t 
by means of Low pass filter with gain Ts

pt     t  nT 
n  
s
Cut off frequency greater than wM

x f t 


xt  x p t 
H  jw
X p  jw 
X  jw
1
.. .
 wM0 wM w
X p  jw 
1 Ts ws  2wM
. 2.w. . w. . . .0 w. .w. . .2.w.
s s M s s w
 ws
 
H  jw
wM  wc  ws  wM  Ts w
Ts
He jw
 2
 wc
. .0 w. 
X f  jw 
c w
0 otherwise
. .
wM 0 wM w Lecture # 5 24
 3.5 SAMPLING THEOREM
Let xt  be a band –limited signal with X  jw  0 for w  wM .
Then xt  Is uniquely determined by its samples xnTs , n  0,  1,  2,......if

Where, ws  2wM and ws  2

Ts
Given these samples, we can reconstruct x(t) by generating a
periodic impulse train.

The successive impulses have amplitudes that are

successive sample values.

This impulse train is then processed through an ideal low

pass filter with gain Ts and cut off frequency greater than ws
and less than ws –wM.

The resulting output signal will exactly equal x(t).

Lecture # 5 25